Spaces:
Sleeping
Sleeping
Create main.py
Browse files
main.py
ADDED
@@ -0,0 +1,76 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# Import libraries
|
2 |
+
import whisper
|
3 |
+
import os
|
4 |
+
import gradio as gr
|
5 |
+
from groq import Groq
|
6 |
+
from gtts import gTTS
|
7 |
+
import traceback
|
8 |
+
from google.colab import userdata
|
9 |
+
|
10 |
+
# Step 1: Load Whisper Model for Transcription
|
11 |
+
try:
|
12 |
+
model = whisper.load_model("base")
|
13 |
+
except Exception as e:
|
14 |
+
print("Error loading Whisper model:", e)
|
15 |
+
model = None
|
16 |
+
|
17 |
+
# Step 2: Initialize Groq Client
|
18 |
+
try:
|
19 |
+
client = Groq(api_key = os.getenv("MY_API_KEY"))
|
20 |
+
except Exception as e:
|
21 |
+
print("Error initializing Groq client:", e)
|
22 |
+
client = None
|
23 |
+
|
24 |
+
# Function to get response from Groq API using LLaMA model
|
25 |
+
def get_response_from_groq(user_input):
|
26 |
+
try:
|
27 |
+
chat_completion = client.chat.completions.create(
|
28 |
+
messages=[{"role": "user", "content": user_input}],
|
29 |
+
model="llama3-8b-8192",
|
30 |
+
)
|
31 |
+
return chat_completion.choices[0].message.content
|
32 |
+
except Exception as e:
|
33 |
+
print("Error getting response from Groq API:", e)
|
34 |
+
return "Sorry, I couldn't generate a response at this time."
|
35 |
+
|
36 |
+
# Step 3: Function to convert text to speech using gTTS
|
37 |
+
def text_to_speech(text):
|
38 |
+
try:
|
39 |
+
tts = gTTS(text)
|
40 |
+
tts.save("response.mp3")
|
41 |
+
return "response.mp3"
|
42 |
+
except Exception as e:
|
43 |
+
print("Error converting text to speech:", e)
|
44 |
+
return None
|
45 |
+
|
46 |
+
# Step 4: Define the Chatbot Function for Gradio
|
47 |
+
def chatbot(audio_input):
|
48 |
+
try:
|
49 |
+
if model is None:
|
50 |
+
return "Whisper model not loaded.", "Unable to transcribe.", None
|
51 |
+
|
52 |
+
# Transcribe audio input using Whisper
|
53 |
+
transcription = model.transcribe(audio_input)["text"]
|
54 |
+
|
55 |
+
# Get response from Groq API using LLaMA model
|
56 |
+
response = get_response_from_groq(transcription)
|
57 |
+
|
58 |
+
# Convert response to speech
|
59 |
+
speech_file = text_to_speech(response)
|
60 |
+
|
61 |
+
return transcription, response, speech_file
|
62 |
+
except Exception as e:
|
63 |
+
print("Error in chatbot function:", e)
|
64 |
+
traceback.print_exc()
|
65 |
+
return "Error occurred during processing.", "Please try again.", None
|
66 |
+
|
67 |
+
# Step 5: Create Gradio Interface
|
68 |
+
try:
|
69 |
+
gr.Interface(
|
70 |
+
fn=chatbot,
|
71 |
+
inputs=gr.Audio(type="filepath"),
|
72 |
+
outputs=["text", "text", "audio"],
|
73 |
+
live=True
|
74 |
+
).launch()
|
75 |
+
except Exception as e:
|
76 |
+
print("Error launching Gradio interface:", e)
|